About TextDB's keywords and tags

[daisynickelsson-code]

Hi and thank you LostRuins for your enthousiasm.

Unless I'm mistaken, I noticed in koboldAILite's JS you only rely on [doc title]. I assume (?) this comes from MiniSearch.
klite.embd.Line 28166 in my version => "paragraphs.push({snippet: currentSnippet, document: title});"
So I have a question for you. Do you plan to handle TextDb on keywords and tags in some [unknown] future?

You can flag me as a "ChatGPT refugee", therefore I'm not very interested in adventure/story modes, but for hybrid assist/chat I found TextDb very reliable to give the AI memory (far more than WI). I've some plan on my side, including modifying your js functions to add a dynamic memory allocator/replacer to maintain consistency over 9 mem slots, right in TextDB.

Now, I'm sure I need those keywords and tags whatever their format, like [document_title]: keyword_1|keyword_2|keyword_3 > tag,tag,tag,tag \n

Do you think you'll add something similar or is this not scheluded and won't in any event?

Thank you
Regards

[LostRuins]

TextDB doesn't just look for the title or tags. It performs similarity matching on the text itself. The retrieved matches consider all content in the snippet, and if you use an embedding model, includes the embedding content as well.